User trust relationship prediction method and system based on graph self-encoding network
A technology of self-encoding network and trust relationship, which is applied in the field of user trust relationship prediction based on graph self-encoding network, can solve the problems that directed symbolic network cannot be directly applied, cannot be effectively processed, and cannot learn negative relations, etc., to achieve accurate network embedding The effect of the result
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Example Embodiment
[0053] Example one
[0054] In order to solve the problem of applying the graph convolutional network to the directed symbolic network, this embodiment first defines the symbolic network adjacency matrix, defines the balance theory, and the form of the propagation adjacency matrix and the directed activation propagation adjacency matrix based on the balance theory. The concept of GCN is extended to the directed symbol network, which describes the basic rules of symbol propagation in GCN. However, the effect of system prediction has not been significantly improved. Experiments show that the graph convolutional network has not learned a lot of effective information from the input matrix. The reason is that for the coding layer-the input information of the graph self-encoding network, that is, the directional activation propagation adjacency The matrix is too sparse (the density of 0 in the matrix is very high), which leads to insufficient prediction basis and limits the accurac...
Example Embodiment
[0133] Example two
[0134] The purpose of this embodiment is to provide a user trust relationship prediction system based on a graph self-encoding network, including:
[0135] Symbol network acquisition module, to acquire comment interaction data between users, and build a user trust relationship network;
[0136] A symbol network processing module, which extracts an adjacency matrix based on the user trust relationship network, and converts the adjacency matrix into a directional activation propagation adjacency matrix;
[0137] Reachability matrix calculation module, combined with symbol network activation and propagation adjacency matrix, calculates symbol network reachability matrix;
[0138] The reachable matrix recursive module combines symbol network activation and propagation adjacency matrix to calculate symbol network reachable matrix, and recursive high-order symbol network reachable matrix;
[0139] The network embedding module takes the high-order symbol network reachabilit...
Example Embodiment
[0141] Example three
[0142] The purpose of this embodiment is to provide a computer-readable storage medium in which a plurality of instructions are stored, and the instructions are suitable for being loaded and executed by a processor of a terminal device:
[0143] Obtain comment interaction data between users and build a user trust relationship network;
[0144] Extracting an adjacency matrix based on the user trust relationship network, and transforming the adjacency matrix into a directional activation propagation adjacency matrix;
[0145] Combine symbol network activation and propagation adjacency matrix, calculate symbol network reachability matrix, and recursive high-order symbol network reachability matrix;
[0146] Use the reachable matrix of the high-order symbol network as the input of the graph convolution network, and use the spectral domain graph convolution method to encode the symbol network to obtain the network embedding result;
[0147] Based on the network embeddin...
PUM
Abstract
Description
Claims
Application Information
- R&D Engineer
- R&D Manager
- IP Professional
- Industry Leading Data Capabilities
- Powerful AI technology
- Patent DNA Extraction
Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.
© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap